Quote:
“Rules should be possible to be understood by those with only modest cognitive capabilities. Theoretically, lawyers should play to role of “explainers” of the rules we call laws. So, I propose lawyer AIs for everyone”.
By Radivis.
There are other posts dealing with AIs in the use of recommender systems, can we use the same base software to another underlying necessity?
I’ve already expressed my opinion to do use of recommender systems for orientation and education in complex systems as the legal system, and the aforementioned quote says it all. And they can be used for decision making processes like voting.
Yes, it seems that there is a lack of serious consideration about how AI may support our decision making processes in the future. There’s a chapter in the book “Anticipating tomorrow’s politics” which for example has quite nice ideas about the use of supporting AIs in teaching. We might want to collect ideas like that for every area of life. These could then be used to inspire both business ideas and Transhumanist Party policies.
If the future is to be good, then everyone should have access to a swarm of really good AI advisors for any field of expertise. The AI advisor swarm might even be integrated or mediated through a kind of warm mentor persona. That would be a real Google-killer. Instead of googling something, you simply ask your AI mentor and get very meaningful answers almost instantly. It seems like Google, Apple, and Microsoft are working on such AI helpers, but they are still far away from being as competent as Watson, for example.
Much simpler systems could rely on the version-nesting wiki (you know, the one I talk a lot). This one can add an agent-oriented background through human computation, solving partially the need for programming.
Done by bringing results from the database of “ideas based on ideas” made by humans themselves, we could feed technologies like swarms with it, adjusting it to individual configurations.
As for the Intelligent Personal Assistant, there is an open source one we could use:
One thing to remark, is advisable to build a common base between this “lawyer AI” and technologies like blockchain to save effort, when possible of course.
The SIRIUS Intelligent Personal Assistant you’ve linked there looks really promising. I’m glad to see that there’s a real open source effort for AI assistants.
And yes, using versioned wikis instead of Wikipedia makes a lot of sense. At least once there versioned wikis are actually well developed and established. The users would then have to specify which versions they prefer, and the AI assistant may suggest that there are other interesting versions of whatever information is requested. This may even be relevant to the law sector. While there is only one official version of a law, there might be different unofficial interpretations of what they mean in certain cases. The lawyer AI might then make predictions about outcomes of certain hypothetical lawsuits.
The main problem is where the data for preferences are stored. It will be the most confidential data a person has. It has to be owned and access controlled by each individual because the AI would have to utilize a data mining approach which could include ambient intelligence, ubiquitous computing, Internet of Things, etc.
IBM has a promising approach with ADEPT. SAFE Network has a promising approach as well. Tau has an interesting approach which is very ambitious and could be promising.
SIRIUS is interesting. If it can be plugged into decentralized data storage and if you can access the necessary computing power to do cool things then it could be great. For data storage I cannot speak for anyone else but I wouldn’t want my private thoughts, preferences, the state of my refrigerator and washing machine to be freely given. I want to control that data and who can access it. I want to be able to sell it to advertisers or keep it to myself.
